一种基于数据垂直划分的分布式密度聚类算法.pdf

IsSN 计‘算机研究与发展 1000—1239／CN11—1777／TP of Researchand Journal Computer Development 一种基于数据垂直划分的分布式密度聚类算法 倪巍伟1 陈 耿2 孙志挥1 1(东南大学计算机科学与工程学院南京2l0096) 2(南京审计学院审计信息工程实验室南京210029) (wni@seu．edu．cn) AnEfficient for Density-BasedClusteringA190rithmVertically DistributedDatasets Ni Weiweil，ChenSunZhihuil Gen92，and 1(DPpnr￡2e起￡0．，^CD，”户“fPrSci已，zce以，z矗E”gi咒韶r西29，SD乱￡^∞s￡8r咒U，zi勘P，百i砂，Nn，巧i扎g210096) 210029) 2(Ln60m￡oryo，A“d赴J理如rmⅡ￡io，2E九gi卵卵ri规g，M巧i，29A“d矗U挖i伽i砂，N口巧i挖g Abstract isan researchindata massiVedatasetshas Clusteringimportant mining．Clustering especially beena forits scaleand Distributed isan t∞muchnoisedata effective challenge1arge points． clustering of methodtoSolvethese Most distributed researchaimsatcircumstancesd pr。blems． existing clustering dataset．Inthis distributed horizontallypartitioned paper，consideringverticallypartitioned onthe of local relationsbetweennoisedatasetsandthe anefficient analysis correspondingglobaIone， is tothe can eliminatethe affectionofn。isedataand filtering noise，whichefficiently negative applied gl。bal reduce of to onthe thescaledatasetbedealt centernode．Furthermore，aneffect structureCTL storage is tostoretheintermediateresultsofeach whichca